Three essays on credit risk modelling for practitioners and supervisors

Since the financial crisis, authorities are putting in place a set of reform measures to strengthen the regulation, supervision and risk management of the banking sector. In response to the complexity of financial instruments as well as to the complexity of the sources of economic and financial shocks, a key feature of these measures is the requirement to strengthen the risk management methodologies of financial institutions. The aim is to ensure a sound management and coverage of the risks to which institutions are or might be exposed and of risks institutions may pose to the financial system. Credit risk is one of the major risks for banks, because it arises from different business lines and has a major impact on capital and solvency. As a result, over the last decade all banks have improved, at different extent according to their size, the quantitative processes to model credit risk. Design, development, implementation and validation of credit risk models represent fundamental components of the Supervisory oversight of banking institutions. At the same time, the output of such models plays and important role in the internal processes of risk management, portfolio management, capital allocation, pricing and planning. The three essays of the thesis join this framework by addressing theoretical and practical issues that Practitioners and Supervisors face when it comes to model and manage credit risk in the contest of bond portfolios and loan portfolios. Subject of the first and second essay is the credit risk in the Euro-zone Sovereign bond market. The first essay proposes an econometric model that Practitioners could use in their portfolio management operations, to monitor and exploit some potential market inefficiencies to price credit risk between the bond market and the related derivative market. In other words, the inefficiencies in the price discovery process of credit risk. The empirical evidences obtained in periods characterized by different levels of credit risk, and then associated to different volatility and liquidity conditions, suggest implications also for Policymakers interested in maintaining the stability of financial systems through timely and correct responses to possible price shocks. Moreover, it arises that limited attention has been devoted in literature to use price discovery models for policy analysis. These findings represent the research premise for the second essay to address, in the same contest of price discovery, the Supervisors’ point of view. The second essay proposes an econometric approach for analysing the effects of monetary policies, of bailout policies and of regulatory innovations on the microstructure of financial markets as well as in measuring the effectiveness of policies in reducing the financial instability. A distinctive feature of the model is to introduce the concept of time-varying price discovery in order to adjust current time-invariant price discovery models, so far used in finance theory for studying the microstructure dynamics of financial markets, in a way that are suitable for policy analysis. Indeed, the time-varying technique leaves the model free to estimate when structural changes happen in the price discovery dynamics and, then, to analyse if policy announcements play an important role in these changes. The empirical application of the model, on a topic currently investigated in literature “the impact of Naked CDS Ban on the sovereign credit risk in Euro-zone”, shows the impact of financial regulation as well as of monetary policies and of crisis events on price discovery of credit risk, such that policy implications can be derived. This model could contribute on the literature of both financial market microstructure and policy impact analysis. Subject of the third essay is the credit risk stress test exercise on loan portfolios requested by the European Central Bank to the majority of European Banks. The essay presents a methodological approach, compliant with regulatory requirements and European Banking Authority scenarios, to build a satellite model for stress testing the probability of default of a loan portfolio. The essay underlines that satellite models presented in literature and, then, used by Practitioners, are biased because do not control for the idiosyncratic risk. Indeed, as pointed out by Souza and Feijò (2011) the causes of the risk of defaulting on bank loans can be divided into two groups: macroeconomic (or structural) factors and microeconomic (or idiosyncratic) factors. While the first group is linked to the general state of the economy, the second group is related to the individual behaviour of each bank and its borrowers. A distinguishing feature of the model proposed is to extract these two factors and, then, disentangle the idiosyncratic factor in the Management component and in the Residual component. The first component models the control over idiosyncratic risk from managers used to offset changes in the level of PD. For instance, an increase of macroeconomic risk generates an increase of default correlations across borrowers, but the entity of this increase will vary among banks according to their strategic decisions and managerial ability. The second component is related to the intrinsic features of the borrowers and is not correlated with the systematic factor. The model has implications for Supervisors and credit risk managers of European banks. For Supervisors interested in a common methodology to compare the effects of an economic scenario on different institutions. For credit risk managers interested in an internal model, compliant with ECB stress testing framework, to estimate a point-in-time probability of default, that is a probability of default responsive to forward-looking macroeconomic information. At the same time, this satellite model can be used for other regulatory exercises, such as the ICAAP report and “the expected credit losses impairment” modelling under the new accounting standard IFRS 9. Moreover, the model could be adopted for other activities, different from risk management, such as budget and planning operations.